linux-kernel - Re: [PATCH] ARM: dts: imx6qdl-hummingboard: Add rtc0 and rtc1 aliases to fix hctosys

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8d5bfef0-ced5-4f3f-a176-707b0dcbd0ea@solid-run.com>
Date: Mon, 29 Jan 2024 19:26:07 +0000
From: Josua Mayer <josua@...id-run.com>
To: "Russell King (Oracle)" <linux@...linux.org.uk>
CC: Rob Herring <robh+dt@...nel.org>, Krzysztof Kozlowski
	<krzysztof.kozlowski+dt@...aro.org>, Conor Dooley <conor+dt@...nel.org>,
	Shawn Guo <shawnguo@...nel.org>, Sascha Hauer <s.hauer@...gutronix.de>,
	Pengutronix Kernel Team <kernel@...gutronix.de>, Fabio Estevam
	<festevam@...il.com>, NXP Linux Team <linux-imx@....com>, Yazan Shhady
	<yazan.shhady@...id-run.com>, Jon Nettleton <jon@...id-run.com>,
	"devicetree@...r.kernel.org" <devicetree@...r.kernel.org>,
	"linux-arm-kernel@...ts.infradead.org"
	<linux-arm-kernel@...ts.infradead.org>, "linux-kernel@...r.kernel.org"
	<linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] ARM: dts: imx6qdl-hummingboard: Add rtc0 and rtc1 aliases
 to fix hctosys

Am 19.01.24 um 15:33 schrieb Russell King (Oracle):
> On Fri, Jan 19, 2024 at 01:46:26PM +0000, Josua Mayer wrote:
>> Am 19.01.24 um 13:07 schrieb Josua Mayer:
>>> Am 18.01.24 um 17:07 schrieb Russell King (Oracle):
>>>> On Thu, Jan 18, 2024 at 04:01:10PM +0100, Josua Mayer wrote:
>>>>> HummingBoard has two RTCs, first integrated within SoC that can be used to
>>>>> wake up from sleep - and a second on the carrier board including back-up
>>>>> battery which is intended for keeping time during power-off.
>>>>>
>>>>> Add aliases for both, ensuring that the battery-backed clock is primary
>>>>> rtc and used by default during boot for restoring system time.
>>>> Given that the snvs RTC isn't battery backed, should we even be enabling
>>>> that in DT?
>>> In imx6qdl.dtsi it is not disabled.
>>> According to Jon it is useful because it can wake up the soc from sleep,
>>> whereas the external rtc can't.
>>>> Also, have you seen any issues such as:
>>>>
>>>> [    0.933249] rtc-pcf8523 0-0068: failed to set xtal load capacitance: -11
>>>> [    0.933505] rtc-pcf8523: probe of 0-0068 failed with error -11
>>>>
>>>> which seems to be exhibiting itself on my SolidSense board?
>>> Not on my HummingBoard Gate Rev. 1.4., but indeed on my solidsense
>>> unit too, which is probably same age as yours.
>>> Only tested imx6dl-hummingboard2-emmc-som-v15.dtb,
>>> but solidsense one should make no difference.
>> I was reading control registers 1-3:
>> debian@...imx6:~$ sudo i2cget -y -a -f 0 0x68 0x00
>> 0x00
>> debian@...imx6:~$ sudo i2cget -y -a -f 0 0x68 0x01
>> 0x00
>> debian@...imx6:~$ sudo i2cget -y -a -f 0 0x68 0x02
>> 0x04
>>
>> ^^ This means low voltage on back up battery
> Interesting - in my case, the solidsense has been powered on for months
> (it's my internet router on the boat).
>
> I've rebooted it again today, and this time it seems to have been
> successful, and is using the time from it.
>
>> After a few power-cycles that error went away.
>> Why pcf8523_load_capacitance would ever return EAGAIN I don't see.
>>
>> In any case now that probe succeeded, I read these values:
>> 0x80
>> 0x00
>> 0x04
> For me, after the last reboot, they contain:
> 0x80
> 0x00
> 0x08
>
>> Long story short I don't think the EAGAIN during probe is related
>> to adding aliases.
>> HOWEVER imo pcf8523_probe should return an error when
>> pcf8523_load_capacitance fails.
> I think the real question is where is the EAGAIN coming from and
> why.
I have tried reproducing this on 6.7.0 with imx_v6_v7_defconfig:
I use expect [1] to capture kernel rtc messages during boot,
explicitly overwrite load capacitance bit to ensure
regmap_update_bits in pcf8523_load_capacitance has work
to do during next probe, and finally trigger software reboot.

On HummingBoard-2 I  have not seen the issue during 80
reboot cycles, and on solidsense not seen during 25 cycles.

Maybe it needs an almost-all-modconfig kernel?
Or maybe it only happens when the rtc has lost time?
Perhaps I can try disassembling the backup battery ... .

[1]
#!/usr/bin/expect -f

set tty [lindex $argv 0]
set logfile [lindex $argv 1]
set count 0

# log full output to file, but only status messages to stdout
if {$logfile != ""} {
    log_file -a $logfile
    log_user 0
}

#exec stty -F $tty 115200 raw -clocal -echo -istrip -hup
#spawn -open [open $tty w+]
spawn tio $tty

while {true} {
    expect {
        -re "Debian GNU/Linux 11 sr-imx6 ttymxc0" {
            set count [expr $count + 1]
            set time [clock seconds]
            set timestr [clock format $time -format "%D %T"]
            send_user "\[$timestr\] Successful boots to Linux: $count\n"
        }
        -re "sr-imx6 login:" {
            send "debian\n"
            expect -re "Password:"
            send "debian\n"
            expect -re {debian@...imx6:~$}
            send "sudo i2cset -f -y 0 0x68 0x00 0x00; sudo reboot\n"
        }
        -re "password for debian:" {
            send "debian\n"
        }
        -re {\] (rtc-pcf8523.*)\n} {
            set matched $expect_out(1,string)
            set time [clock seconds]
            set timestr [clock format $time -format "%D %T"]
            send_user "\[$timestr\] $matched\n"
        }
    }
}